[sdiy] MIDI isn't musical : Flame bait?

[Thomas Hudson]

On Sunday, January 13, 2002, at 07:16 PM, Ian Fritz wrote:

> At 06:33 PM 1/13/2002, Thomas Hudson wrote:
>
>> I think Don's original point is still valid. Guitar synths have to 
>> send six separate channels of MIDI
>> notes, along with copious amounts of pitch bend info.
>
> I don't understand this. Why are six channels (as opposed to six-voice 
> polyphony) required?

Pitch bend is per channel. Thus if I hit an open low E and then bend a 
note on he high E string,
I only want the second note to change pitch.
>
> Six voices and pitchbend doesn't sound to me like any more information 
> than is commonly handled by a poly synth with aftertouch. Pitchbend 
> changes quite slowly.

Not when you're doing finger vibrato.
>
>> Secondly, there is nothing in the protocol
>> to represent many common guitar techniques. You press a key on a 
>> keyboard and it sends the
>> NOTE ON message with note number and velocity, then sends NOTE OFF (or 
>> NOTE ON w/
>> zero velocity) when you release the key. But how do you represent a 
>> pulloff on guitar?
>
> Be careful not to confuse (1) how to generate control information and 
> (2) how to transmit it over MIDI. Within the MIDI protocol there are 
> *plenty* of CC messages available to transmit this kind of information. 
> Making a control device that could sense the pulloff gesture (and 
> others) would certainly be a challenge. Designing a synth to use this 
> information effectively would be another challenge. But these are not 
> MIDI issues.
>
However, having generalized methods as opposed to specific methods mean 
no synth manufacturer
will implement it the same or at all. Have you ever seen any synth that 
could use a CC as "don't retrigger
on the next incoming note on, but rather do it legato." Unless it is 
specified there is no agreement.

>> MIDI may be good enough for keyboard players, but if falls woefully 
>> short for guitar, sax, trumpet,
>> harmonica, etc. Basically any instrument where the technique of 
>> playing a note involves more
>> than how fast you press a key.
>
> I disagree. For example the Yamaha and Synthophone wind controllers 
> allow a wide range of expression. The Yamaha WX controller, for 
> instance, sends up to 5 control messages simultaneously. Another good 
> place to learn about what MIDI can do is Bouvard Hosticka's 
> Elect-RO-Clar:
>
> http://windsynth.org/iwsa_labs/non_commercial_controllers/Elect-RO-Clar/
>
>
>> Secondly, most of today's multitimbral synths still only have a single 
>> MIDI input port. So while there
>> may be enough bandwidth for a single part, there isn't enough for 16 
>> parts. Of course this isn't
>> so much the protocol as the interface, something like Yamaha's mLan 
>> could solve this.
>
> Again, I'm not following you. Are you saying that 16 MIDI channels 
> simultaneously are needed? Why?

The synth is capable of playing 16 different patches at once. ch 1, 
piano, ch 2 drums, ch 3 bass,
ch 4 guitar, etc. Imagine 16 different synths fed by a single MIDI pipe.


>
>> But if even if someone came up with the theoretically perfect guitar 
>> to MIDI converter, and resurrected
>> Hendrix from the dead (the former probably being more difficult) to 
>> play it, what would get recorded
>> into a MIDI sequencer could in no way equal what was played.
>
> What do you base this statement on? Human beings can only produce and 
> process information at a finite rate. Exactly why don't you think MIDI 
> can handle this rate?

It's not rate. It's nuance. Pick up a guitar and scrape the pick on the 
strings. Do different types of scrapes.
Turn the volume up and hit the body of the guitar. Drop it. Press down 
and hold the whammy while shaking
the guitar to slap the loose strings against the pickups. Take the pick 
and press down on the strings
between two pickups, scraping the strings on the magnets. Grab a small 
television set and move it over
the pickups while the guitar is feeding back. Turn and run into your 
stack of Marshall amps, humping and
grinding the guitar against the amps.

Now sit down and write out the spec for a protocol that would capture 
all this.

Oh wait, now I want to use an eBow. Your protocol is out of date. And 
I've left out all the more conventional
things guitarist might do like false harmonics, relying on the changing 
of tone by different pick positions,
the difference between finger picking and using a plectrum.

The difference between the analog world and the digital world is that 
analog responds in interesting
ways to the unpredicted, digital usually crashes :-). Protocols can only 
represent was has been
predicted.

I'm very familiar with the MIDI spec and agree that there is much more 
there than has been used
by manufacturers. Heck, most of them still haven't produced synths that 
allow you to specify 14
bit CCs for parameter control.

>
>> And it's why you'll never find a MIDI file on the web of Hendix' "Star 
>> Spangled Banner" or
>> "Machine Gun" that would even fool a non-guitarist.
>
> Do you mean not now or not ever. Well, people made similar statements 
> about how a computer could never play a world-class game of chess. But 
> it happened.
>
What Harry said. Not with the MIDI protocol, maybe some other.

But there are so many nuances to playing guitar that the protocol would 
be complex.

Tomy